摘要
本研究采纳了先进的基于卷积神经网络(CNN)的姿态估计技术研发的一套太极拳训练与教学系统,命名为GT-TaiJi系统。为了提高GT-TaiJi系统智能评分指标误差阈值设置的适用性。本文以太极拳专业评价要求为理论依据,通过对智能评分与专业裁判员评分进行比较和相关性分析,观察智能评分指标误差阈值设置对智能评分与专业裁判员评分一致性的影响,旨在提高系统智能评分的可信度和专业性。研究结果显示,评分指标误差阈值设置优化前,智能评分与专业裁判员评分结果呈非常显著性差异(r=0.79;p<0.01);优化后智能评分与专业裁判员评分结果无显著性差异(r=0.81;p>0.05)。有效提高了人工裁判员评分与智能评分的一致性,弥补了系统智能评价指标模糊性的不足。
In this study,a Taijiquan training and teaching system was developed based on advanced convolutional neural network(CNN)-based pose estimation technology,named GT-TaiJi system.In order to improve the applicability of the error threshold setting of the intelligent scoring index of the GT-TaiJi system.Based on the professional evaluation requirements of Taijiquan,this paper compares and analyzes the correlation between the scores of the intelligent score and the scores of the professional referees,and observes the influence of the error threshold setting of the intelligent scoring index on the consistency of the scores of the intelligent scores and the scores of the professional referees,so as to improve the credibility and professionalism of the system's intelligent scores.The results showed that there was a significant difference between the scores of intelligent scoring and the scoring results of professional referees before the optimization of the error threshold setting of scoring indicators(R=0.79;P<0.01),and there was no significant difference between the scoring results of intelligent scoring and professional referees after optimization(R=0.81;P>0.05).It effectively improves the consistency between the manual referee score and the intelligent score,and makes up for the lack of ambiguity of the system's intelligent evaluation index.
作者
周丽娟
胡茂荷
郭明福
ZHOU Li-juan;HU Mao-he;GUO Ming-fu(Guangzhou University of Physical Education,Guangzhou 510500,China)
出处
《福建体育科技》
2025年第2期82-86,共5页
Fujian Sports Science and Technology
基金
2022年广东省高校科研平台和科研项目(重点领域专项(自科))(项目编号:723018175)。
关键词
太极拳
姿势评估系统
评价指标误差阈值
优化
Tai Chi
postural assessment system
evaluation index error threshold
optimize
